At Google Cloud Next '23, held at the Moscone Center in San Francisco, Google mainly continued the focus on artificial intelligence technology applications during Google I/O 2023 this year, thereby strengthening the scope of Google Cloud's technical cooperation and integrating its technologies with many industries to promote the development of artificial intelligence applications.
New Cloud TPU v5e accelerator, partnering with NVIDIA to accelerate artificial intelligence computing
Google announced the release of its new Cloud TPU v5e accelerator, designed to meet the training and inference acceleration needs of small and medium-sized businesses for artificial intelligence applications, while offering improved cost-performance. Additionally, Google will launch the A100 cloud virtual machine, powered by NVIDIA H3 GPUs, enabling businesses to meet the computational demands of large-scale natural language models.
In response to the current trend of multi-cloud applications, Google has announced Cross-Cloud Network, enabling Google Cloud customers to connect various applications across cloud environments while ensuring data privacy and security. Through the Google Distributed Cloud architecture, Google Cloud customers can run various artificial intelligence and data processing processes using Vertex AI and AlloyDB Omni services.
In the update of Google Kubernetes Engine (GKE) managed containerization service, Google launched the enterprise version of GKE service, making it easier for enterprises to build multi-cluster, horizontally scalable artificial intelligence and machine learning operations.
Upgrade the large-scale natural language model PaLM and expand it to support more open source model frameworks
Google has also upgraded its large-scale natural language model, PaLM, by increasing its input parameter length by four times and adding support for 4 languages. Previews are already underway for interfaces in over 38 languages. Codey, a tool for developers that automatically generates code, has improved content generation accuracy by 100% and is compatible with the most popular languages in the market. Imagen, an automatic image generation service, has enhanced its functionality and added a Style Tuning feature, allowing brands to quickly generate 25 reference images for use in branding, product design, and more.
As for expanding Vertex AI's capabilities, Google has added Meta-supported open-source model LLaMA 2 to its model database; Claude 2 from Anthropic, an AI startup founded by former OpenAI employees; and Falcon, an open-source AI model from the UAE government-funded Institute for Technology Innovation. These additions allow Google Cloud users to more easily access these model resources and quickly build AI application services.
Collaborate with more industry players to expand the application of artificial intelligence
At the Google Cloud Next '23 event, Google announced technical collaborations with Tel Aviv-based artificial intelligence (AI) technology company AI21 Labs, Bayer Laboratories, employee experience platform provider Culture Amp, Fox Sports, GE Appliances, General Motors, American Hospital Corporation, IHOP, the largest breakfast waffle chain in the United States, artificial intelligence startup Runway AI, Meditech, Morgan Stanley Capital, Six Flags, the world's largest theme park chain, and The Knot, an online wedding planning service platform.
At the same time, Google Cloud has integrated its AI technology with SAP and digital document signing service DocuSign, and has established in-depth collaborations with data marketing company Acxiom, Bloomberg News, data analytics company CoreLogic, consumer credit investigation company Equifax, consumer research and retail data monitoring agency NIQ (Nielsen AQ), which recently completed its merger with GfK, and credit monitoring service provider TransUnion, to build and train various AI technology application services through the Google Cloud platform.
